SHALLOW WATER ACOUSTIC AMPLITUDE
FLUCTUATIONS AT 35 AND 65 KHz
Michael Thomas Korbet
L
II
L
onterey, California
SHALLOW WATER ACOUSTIC AMPLITUDE
FLUCTUATIONS AT 35 AND 65 KHz
by
Michael Thomas Korbet
Thesis Advisor:
W. Denner
March 197^
Approved Ion public KeJL&a&e.; dU>t>uhation anlimLtad.
T160860
Shallow Water Acoustic Amplitude Fluctuations
at 35 and 65 kHz
by
Michael Thomas JCorbet
Lieutenant, United States Navy
B.S., United States Naval Academy, 1966
Submitted in partial fulfillment of the
requirements for the degree of
MASTER OF SCIENCE IN OCEANOGRAPHY
from the
NAVAL POSTGRADUATE SCHOOL
March 1974
c /
ABSTRACT
An underwater acoustics experiment conducted in shallow
water (70 feet) off the New Zealand east coast in 1972-1973
is described. Short acoustic pulses of 35 and 65 kHz sound
were projected along near-orthogonal paths of approximately
300 yards. Environmental parameters were simultaneously
observed.
Statistical and spectral analyses of pulse heights were
performed on 12 selected runs using digital techniques.
Coefficients of variation ranged from 2.0% to 15.5%. In
almost all cases, higher variability was observed along the
acoustic path oriented perpendicular to the predominant
swell direction. Along this same path, periods corresponding
to common surface swell periods were frequently evident in
the autocorrelation functions of the fluctuations. Coherence
between the fluctuations along each path was low, averaging
about 0.1. Long period oscillations suggestive of modulation
by internal waves were apparent in several runs. No signif-
icant dependence of variability on acoustic frequency was
detected.
Microscale temperature fluctuations measured simultane-
ously are discussed.
TABLE OF CONTENTS
I. THE EXPERIMENT 8
A. BACKGROUND 8
B. OVERVIEW 12
C. PHYSICAL DESCRIPTION 13
D. DESCRIPTION OF SENSORS 21
1. General 21
2. Acoustic System 22
3. Temperature System 26
E. DATA ACQUISITION 27
II. DATA ANALYSIS 31
A. STRIP CHART RECORDING ■ 31
B. DIGITIZATION 31
C. CONVERSION 35
D. PULSE HEIGHT MEASUREMENT 35
E. SELECTION OF RUNS FOR DETAILED ANALYSIS 36
F. STATISTICAL ANALYSIS 39
G. SPECTRAL ANALYSIS 42
H. DISPLAY OF RESULTS 44
III. RESULTS OF ANALYSIS 45
A. TEMPERATURE FLUCTUATIONS 45
B. STATISTICAL RESULTS OF PULSE HEIGHT
ANALYSIS 46
C. SPECTRAL ANALYSIS RESULTS 48
D. DISCUSSION 52
IV. CRITIQUE OF EXPERIMENT 62
A. PARAMETERS 62
B. RUN LENGTH 63
C. RECORDING OF DATA 64
D. PHASE II RUNS 65
E. RELFECTED SIGNALS 66
F. LOG KEEPING 67
V. SUMMARY, RECOMMENDATIONS, AND CONCLUSIONS 68
APPENDIX A: PULSE HEIGHT TIME SERIES AND
ANALYSIS RESULTS 72
LIST OF REFERENCES 125
INITIAL DISTRIBUTION LIST 12 8
FORM DD 1473 130
LIST OF TABLES
I. Basic Data on Analyzed Runs 37
II. Sea and Weather Conditions During Each Run 38
III. Summary of Statistical Analysis Results 47
IV. Summary of Selected Results from
Spectral Analysis 49
LIST OF FIGURES
1. Map of Auckland-Leigh area in New Zealand 15
2. Map of experiment site near Leigh, New Zealand 16
3. Schematic of acoustic propagation equipment 23
4. a. Typical sample of microscale temperature
fluctuations and simultaneously observed
acoustic pulse heights 32
4.D. Expanded view of acoustic pulse sequence 32
5. Graph of coefficient of variation versus
number of bad pulses 53
6. Graph of bad pulses on Hydrophone 1 versus
those on Hydrophone 2 5^
7. Histogram from Pulse Height Analyzer for
pulses recorded during the 5 minute period
shown on temperature record: low thermal
activity 56
8. Histogram from Pulse Height Analyzer for
pulses recorded during the 5 minute period
shown on temperature record: high thermal
activity ■ 57
ACKNOWLEDGEMENTS
The author wishes to acknowledge the valuable guidance
and encouragement of Dr. Warren W. Denner from whom,
through countless hours of stimulating dialogue, I learned
the value of innovative and exploratory thinking. The
assistance and comments of Dr. Edward B. Thornton and Dr.
Robert H. Bourke are greatfully appreciated.
Particular note is due the Department of Physics of the
University of Auckland for the cooperation extended during
the conduct of the experiment. Special gratitude is offered
to the staff of the Naval Arctic Research Laboratory,
especially Mrs. Dorothy Underwood, for the gracious hospi-
tality and multiple resources made available during the
author's work at that facility.
The experimental research discussed in this thesis was
supported through contracts from the Naval Ordnance Systems
Command (Code 03C) .
I. THE EXPERIMENT
A. BACKGROUND
Acoustic fluctuations in the ocean first gained serious
attention as a result of sonar development and acoustic
experimentation during and immediately after World War II.
Variability in the pressure amplitude of an acoustic signal
received from a constant source was attributed to the
presence of moving inhomogeneities which continuously altered
the ray paths for energy to travel from the source to" the
receiver.
During this same time frame, the development of quick-
response thermopiles enabled Urick and Searfoss (1948, 19^9 )
to measure the micro-scale thermal structure of the ocean
near Key West, Florida. Recognizing the prominent role of
temperature in affecting the acoustic refractive index, they
proposed that the "thermal patches" evidenced in their
experiments were the cause of acoustic fluctuations.
Subsequent investigations — theoretical, laboratory, and
in situ — are summarized through 1964 by Urick (1967) • A
more detailed theoretical development and analysis can be
found in Skudryzk (1963).
Although a clearer understanding of the nature of the
ocean'.s fine structure is available now, compared to twenty-
five years ago, the equations often used to predict acoustic
variability are still those developed by Bergmann (1946) and
Mintzer (1953a, 1953b, 1954). Under the condition that
r << ka
-— where r = range from source to receiver,
k = acoustic wavenumber,
a = thermal microstructure scale length, i.e.,
the "patch size",
the acoustic variability, calculated as the coefficient of
variation, V, is given by:
*L Js U2r3 jJs
V " ( 15 * a3
where u is the RMS refractive index contrast of the
inhomogeneity .
2
At relatively long range, such that r >> ka , the
variability is given by:
„ _ t IT'S 2, 2 ,h
V = ( -75— y k ar)2
These two equations were derived under the assumption of
a Gaussian autocorrelation function for the temperature
fluctuations. The quantities V and u are calculated as:
V =
<(P -
)^>
(
)2
coefficient of variation; i.e., V is the
fractional standard deviation of the pressure
amplitude, P
V =
(1)
, where c = sound speed
= RMS variation of the refractive index.
The "patch size", a , is determined from the space auto-
correlation function of the temperature fluctuations as
measured by a sensor moving along a line through the tempera-
ture field. Invoking the hypothesis of "frozen turbulence,"
and allowing the temperature field to be advected past a
fixed sensor, • "patch size" can also be determined from the
temporal autocorrelation function of the temperature fluctu-
ations, (t), where x is the lag time, provided the advection
velocity is known.
Mintzer (195*0 has shown that the autocorrelation func-
tion for the acoustic fluctuations, R(x), should be related
to the autocorrelation function for the temperature fluctu-
ations by the equation
to
o
Ul
Z
o
h-
X
T-
V>
0.
O
o
O
tc
Z
<
a
u
>
o
X
m
£
S
D,
•H
<1J
d
o
23
second oscillator which produced the pulse. The second
oscillator, in turn, triggered the power amplifier and
simultaneously input the trigger signal to a specially
designed sample and hold circuit. The power amplifier drove
a horizontally omnidirectional projector at one of the two
experimental frequencies, 35 kHz or 65 kHz. The signals
from the hydrophones were passed via cabling back to the
shore station where they were input to the dual channel
sample and hold circuit .
This circuit, designed and constructed by personnel
of the Physics Department at the University of Auckland in
New Zealand (Ash 1972) operated as follows. The trigger
signal from the second oscillator initiated a sample inhibit
instruction in the sample and hold circuit to prevent the
circuit from being falsely triggered while the acoustic pulse
was traveling down range. The pulse output from the hydro-
phone preamplifier, approximately 2 volts peak-to-peak, was
amplified in the circuit to approximately 8 volts peak-to-
peak. By the time the pulse arrived at the circuit, the
sample inhibit signal had been cancelled. The first cycle
from the arriving pulse that exceeded a pre-set comparator
level enabled a 200 msec sample hold component. At this
point the peak-to-peak voltage value of the pulse cycle was
held for 200 msec. The 200 msec output signal was then
applied to a tape recorder. Approximately five cycles of
the pulse were required to initiate the sample and hold
process. After a pulse was Identified, the sampling circuit
2 'J
was disabled by a logic sub-circuit to avoid triggering of
the system by noise or reflections between direct path pulse
arrivals .
Pulse durations of 1.0, 1.5, and 2.0 msec were used
in various phases of the experiment. However, this did not
affect the operation of the sample and hold circuit since
pulse height was determined in the first five cycles of the
pulse train and later arrivals were ignored.
Pulse repetition rate as determined by the first
oscillator also varied, with nominal rates of 1 and 3 pulses
per second in use at different times. Problems associated
with the higher rate will be discussed in Chapter IV.
■The three transducers used in the experiment were
basically identical. Each transducer was comprised of a low
frequency and high frequency section, resonant to 35 kHz and
65 kHz respectively. The transducers were Model ITC601D
manufactured by the International Transducer Corporation of
Goleta, California.
Peripheral equipment consisted of an oscilloscope
which was used to monitor the performance of the sample and
hold circuit and to observe the degree of fluctuation
occurring during a run. In the latter stages of the experi-
ment, a pulse height analyzer was also placed in line to
obtain immediate analysis of height distributions.
Range geometry and water depths were such that
surface-reflected and bottom-reflected signals arrived at the
hydrophones sufficiently late enough to Insure that only
25
direct path measurements were made. During the initial
phase of the experiment, however, occasional interference
from surface-reflected arrivals did occur at the Bird Cage
Site under conditions of low tide and an acoustic frequency
of 35 kHz. A satisfactory solution to this problem was
obtained by lowering the mast at the site by 5 feet prior to
the commencement of later phases.
3. Temperature System
Glass encapsulated, fast response thermistors were
used for measurement of gross temperature and temperature
microstructure . For the gross temperature thermistor, the
Wheatstone bridge components were chosen so that the bridge
balanced at l4.5°C, the center of the anticipated temperature
range of 12°C to 17°C. A temperature range of this magnitude
was expected due to the long period of time planned for the
conduct of the experiment.
With the high resolution required for the micro-
structure measurements, a small change in gross temperature
would have quickly moved the microstructure bridge off
balance if a conventional Wheatstone bridge arrangement were
used. Thus, high resolution would have been available only
over a very limited temperature range. This problem was
overcome by using two equal resistors in the ratio arms of
the bridge and two matched thermistors in the opposite arms.
Thermal inertia added to one of the thermistors ensured that
only slow temperature changes were sensed. The other therm-
istor in the adjacent arm sensed both gross temperature and
26
temperature microstructure . Thus, the bridge output was
proportional to microscale temperature fluctuations but had
the ability to stay balanced over a wide range of gross
temperature. The stock of thermistors available was
calibrated in a bath from 11°C to 25°C and proved to be very
well matched and therefore suitable for this type of
arrangement .
For the microstructure measurements, a temperature
change of 0.1°C was set equivalent to a 1.0 volt change in
the bridge output; the microstructure thermistors were
essentially linear over a range of ± 0.1°C from the zero
point. The gross temperature range of 12°C to 17°C was
equivalent to a voltage output range of -2.0 volts to +2.0
volts with 0.0 volts registered at l4.5°C.
To make efficient use of the limited conducting
cables available, temperature data signals were frequency
multiplexed in the' underwater equipment and then transmitted
to the shore equipment in the laboratory where the signal was
demultiplexed by filters. The carrier frequencies for the
two signals were spaced sufficiently far apart to insure
that rejection of the unwanted signal by the filters was
effective .
E. DATA ACQUISITION
Preparation for the experiment began during the summer
of 1971. This included renovation of the Tripod Site and
underwater surveys to determine the best locations for
27
additional instrument masts and optimum routing to bring
the armored cables ashore. Placement of cables, masts, and
instruments was executed at various times through the summer
of 1972.
Phase I of the experiment with all sensors operating was
conducted during the period 26 September to 19 October 1972.
Following a six week break when alterations to the mast at
the Bird Cage Site were made, Phase II was executed in the
first two weeks of December 1972. Phase III of the experiment
took place during the first week of July 1973. The runs
analyzed in this thesis were all made during Phase I.
In order to establish a comprehensive data base, it was
planned to take measurements on an hourly basis and whenever
significant phenomena were observed. During Phase I, this
schedule was put into effect. However, data storage require-
ments dictated that the original schedule be modified to
acquisition of acoustic and environmental data four times
daily and during unusual occurrences (e.g., high swell, high
winds, abnormally calm sea, etc.). The four periods specified
were sunrise, midday, sunset, and midnight. This schedule
was maintained during the latter stages of Phase I and during
Phase II. The schedule for Phase III was further modified
to provide for as near-continuous data collection as possible
during the two days that spanned this phase.
Because of the expansive data accumulation anticipated,
run length during Phases I and II was planned to be nominally
10 minutes. It was calculated that a run of this relatively
28
short length would be long enough to satisfy the requirements
for statistical and spectral analysis of the fluctuations in
the temperature and acoustic data. On occasions when extreme
variability was observed, run length was extended to twenty
minutes. During Phase III, run length was arbitrary, with
the shortest run being 27 minutes and the longest 77 minutes.
In order to investigate the frequency dependence of the
acoustic fluctuations, runs were generally conducted in
series of two, alternating the high and low frequencies in
each series. Due to the mechanical manipulations required
to change frequency, the two runs in a series were separated
by a period of about 8 to 20 minutes.
Two magnetic tape recorders were in use during Phases I
and II of the experiment. One was a four channel Hewlett-
Packard Model 3960, with a tape capacity of 1800 feet of
1.5 mil tape; the second was a seven channel Ampex FR1300
model with a capacity of 2500 feet of 1.5 mil tape. On the
Ampex tapes mean temperature, temperature microstructure,
pulse heights from both hydrophones, and the current data
from the three ducted current meters were recorded. Wave
height data, turbulent velocity data from the vertical compo-
nent ducted current meter, multiplexed temperature data, and
pulse height data on a time-sharing basis were recorded on
the Hewlett-Packard recorder. The first half of a run was
dedicated to Hydrophone 1 at the Bird Cage Site, and the
second half to Hydrophone 2 at the Reef Site.
29
Aanderaa current meter tapes were processed after
retrieval by displaying the data on a calibrated strip
chart. Except for a few runs in Phase II, ambient noise
was not recorded, but was monitored during each run. Also
in use was a four channel strip chart recorder which dis-
played wave and tide information, gross temperature, and
temperature microstructure data on a real time basis.
Log-keeping involved recording of dates, times, tape
recorder channel allocations, significant wave height (H ,_)
stage of the tide, and a summary of weather conditions.
Wave height and tide information were determined to the
nearest 0.1 foot from the strip chart.
Because the experiment was a joint venture between
personnel from the University of Auckland and from the Naval
Postgraduate School, copying of data tapes was required
after each phase to make the data available to both parties.
Selected tapes from the Ampex and Hewlett-Packard recordings
were transcribed onto one inch, fourteen track magnetic
tapes with the use of a Sangamo 3562 tape recorder. Addi-
tionally, copies of the experiment logs and of some of the
current and wave strip charts were returned to the United
States for analysis.
During Phase III, the Sangamo recorder was used for
initial data acquisition and dubbed tapes were provided to
the University of Auckland.
30
II. DATA ANALYSIS
A. STRIP CHART RECORDING
The first step in the analysis procedure was to survey
the transcribed data to determine qualitatively the nature
of the experimental runs available. Initially, this was
attempted by viewing the data on a dual-trace oscilloscope.
However, only two channels could be viewed simultaneously
and no permanent visible record was produced.
The magnetic data tapes were then played back on the
Sangamo tape recorder with the output displayed on an eight
channel Brush strip-chart recorder. The two tape channels
containing hydrophone information were input directly to the
Brush recorder. Each channel containing temperature data
was first low pass filtered through a Krohn-Hite Model 33^0
filter at 20 Hz to eliminate high frequency noise that may
have contaminated the signal. The filtered data were then
displayed on the Brush recorder chart.
A typical section of pulse height data and temperature
microstructure data is shown in Figure 4(a). Figure 4(b)
shows an expanded view of a typical pulse sequence.
B. DIGITIZATION
After screening the data for noise and run length, thirty-
five runs made during the period 30 September through 13
October 1972, and five runs from the period 2 through 3 July
1973 were selected to be digitized. None of the runs from
31
m
— i
13
— : i • «\
u °
CD
w
o
o
- co
--f~^~-
~Z1
1 — ~
P
-— co
1-1
=± ^
1 P-i
cz
I
i ■ r ■• :■ ■!
.
1
s 1
i"~"
-4 3
cd
CO
CD
co
rH
P.
T
-J-
J, o
d
ui E
.X.
: 1
-3
3
3
3
3
J3
o
ctf
^
•H
a
CO
H
rH
a!
e
CO
•
N
ffi
H
C
cti
.C
P
f-i
cu
p
cti
a)
Ph
bO
. >>
•
rH
(U
P
CO
.G rH
hO 3
•H
p.
rH
CO
Ph
0)
W
bO
•H
bO
•H
CU
Ph
P
P
Cti
U
CU
Xi
C P
o
•H
m
P
•H
•H
P
0)
0)
CO
j an
a)
3
In
a.
CD
IH
bO
32
Phase II of the experiment were chosen. Pour channels of
acoustic and temperature data were digitized. Digitization
was accomplished through the use of a COMCOR 5000 analog
computer, a Scientific Data Systems XDS 9300 digital computer,
and a Naval Postgraduate School analog-to-digital conversion
program. All of the above were available in the Electrical
Engineering Computer Laboratory of the Naval Postgraduate
School .
In the digitization process, the most critical factor to
be specified was the sampling rate. Based on a requirement
to accurately define the pulse height on the digitized
record, a sampling rate equivalent to 256 samples per
channel per digital record was selected. This rate translated
to a sampling time increment of 0.04 seconds between samples.
Since the length of a pulse was 200 msec (0.2 seconds), the
sampling rate insured a minimum of five samples on a normal
pulse height and a maximum of six samples if the first sample
on the pulse happened to occur exactly at the leading edge.
Each hydrophone channel was input directly from the Sangamo
tape recorder to the analog patchboard where the signal was
amplified 25 times.
The Nyquist frequency associated with the sampling rate
was given by the equation,
f = 1
c 2At
where At = sampling time increment.
33
For the increment chosen, 0.04 sec, the Nyquist frequency
was 12.5 Hz. This frequency was above the highest frequency
of significant fluctuations expected in the temperature
microstructure data and was high enough to minimize aliasing.
Temperature data were played into the input of the analog
patchboard where a plus 10 gain was applied. The amplified
signal was then low pass filtered at 50 Hz to eliminate
locally generated 60 Hz noise that was apparent in initial
trial runs. Because tape playback during digitization was
conducted at 16 times real time, the actual temperature data
was, in effect, filtered at 3.125 Hz. The filtered signal
was then input to the patchboard where a second plus 10 gain
was applied. Total amplification of the temperature data
was 100 times. Of several methods evaluated for filtering
and amplifying the temperature data, the above method
appeared to be the most practical and efficient in eliminating
the locally introduced 60 Hz signal.
In addition to the above procedures, biasing- of the
amplified signal was required to keep the signal voltage
within the ± 100 volt range of the analog computer. The
COMCOR 5000 provided bias adjust on each channel.
Each analog run was converted to a digital file of
discrete samples. These samples were written on seven track
magnetic tape in octal base notation. Each file contained
all four channels of digitized data applicable to that run.
Data in a file were written in blocks of 1024 samples called
digital records; each record spanned 10.24 seconds of real
time data.
3'J
C. CONVERSION
Continued analysis was to be performed on the IBM 360/67
digital computer at the Naval Postgraduate School. However,
it was first necesary to make the digitized data compatible
with the IBM 360 system. The conversion process involved
changing the notation from octal base to hexadecimal base,
and rewriting the data on a nine track magnetic tape. The
conversion was made using programs available at the Naval
Postgraduate School computer center and described in a
current technical note (Raney 1973) .
D. PULSE HEIGHT EXTRACTION
Of the 256 samples per channel in one digital record
which represented 10. 24 seconds of data, only about 50 of
the acoustic channel samples actually fell on a pulse height
peak. Furthermore, these 50 samples accounted for only 10
pulses. In order to conserve core storage, it was necessary
to compute and store only the pulse height value, rather than
read into memory an entire file, consisting of an average
58 digital records.
A program was written to read from the tape only the two
channels containing pulse data. When one digital record had
been read in, the pulse heights for each acoustic data channel
were computed and stored. Upon completion, the next suc-
ceeding record was read from the tape and processed. This
sequence was continued until the entire file had been
analyzed.
35
Programming precautions were taken to insure that no
pulse heights were missed on either channel and that only
actual pulse heights were retained; i.e., spurious noise
spikes were omitted.
E. SELECTION OF RUNS FOR DETAILED ANALYSIS
Of the 35 runs in Phase I that were digitized, three were
unacceptable due to excessive random noise that was apparently
acquired during digitizing. From the remaining 32 runs,
twelve were selected for detailed analysis. The selection
criteria were to form a suite of runs that were representative
of all typical environmental conditions, and to choose runs
that for one reason or another (e.g., frequency, time of day,
time sequence) were suitable for comparison with each other.
No runs from Phase III were considered due to low apparent
variability and difficulty with the data resulting from
unexplained interchannel feedback somewhere in the data
acquisition stage.
Pertinent details of the twelve analysis files are shown
in Table I. Because of digitizing requirements, the file
length was always slightly shorter than the actual experiment
run length. Table II summarizes the environmental conditions
present during each run. The comments section of Table II
was synopsized from the original logs kept during the
experiment .
36
U to
-
c-~
O
t~-
C^
VO
CM
CM
en
rH
rH
rH
CM
rH
CO
cvj
CM
rH
in
CM
CM
+5 «
rH
rH
rH
m
LT\
in
in
CM
in
0 &0 G
, '
HCrl
6
B
6
•h a) e
B
B
a
a
a
o
S
e
6
a\
o
e
fe ^-^
-
VO
VO
CM
C—
vo
t—
on
vo
CO
0 CD ^
o
o
rH
o
o
o
O
o
O
rH
t—
O
CO CD £h
o
o
O
VO
vo
vo
VO
VO
C\J
OJ
(M
rH
h 6 o
ON
on
on
ON
on
on
ON
on
O
O
3 -H G -P
^EhH
C jG 3
cd P to
muc
y P -H
3
to -P O
G -H P
3 P
K ra-d
G CD
O fc
• O ?H
CO CD
G t—>
rHVO CO
Cd rH U
G rH -H
< cd
T5 a
G G
o cd G
•H
•
cd «
G
P rH T3
O
cd vo cd
•H
Q P
CO
T) o
to
o c 3
3
•H Cd 'CS
o
co c
CO
cd o o
•H
CQ vo o
T3
^>
a
a
a
a
p
P
P
P
P
P
P
p
CM
0
0
0
0
o
O
o
o
O
O
O
o
0
0 t—
00
CO
CO
CO
O
O
o
o
o
O
o
o
rH
P ON
£>
Cd rH
o
o
o
o
rH
rH
rH
rH
CM
CM
en
on
cd
Q —
on
en
on
en
rH
rH
EH
37
o
o
g
O
O
co
o
^J
.C
o
-p
Cd
3
.a
o
K)
CO
*\
u
> '
O en
o r\J
OJ
'Ci rH
o p
rH d
o
O -H
rH Cd
\ hO
co cd
3
P
13
X3
O
M
0)
-P
CO
2
to
•H
h
X3
3 cu
rH
O
P-i CO
CD
H
o
CO
&
a
5 .H
5 £
u
CD
>
o
CJ
T3
o
rH
CJ
O
H
\
C\J
-p
w
%
•H
rH
r?
(D
>
O
O
oo
p
cd
3
bO
o
bO
P
CO
'r-3
CO
■6
I
rH
CO
P
CO
I
I
co
Cm
O
P
CO
?
■d
CU
p
CO
cd
o
m
u
O
'O
CO
en
cd
'd o
CD M
CJ Cm
B
CO
%.B
Cd °
B
T3
q on
•H vH
3
O CO
VO VO
O O
VO VO
O O
CTi
O
oo
o
t— vo in
o o o
vo
o
co
H
co
LO
OO
m
co
o
o
t~-
oo co
o
+
o
vo
+
rH
+
rH
I
OJ
I
CM
I
C\J
I
OJ
I
H
I
in
vo
OO
cr\
vo
OO
■=r
-=T
-=r
^r
LT\
LA
O
rH
OO
on
VO
VO
VO
t—
t-
rH
rH
38
P. STATISTICAL ANALYSIS
In order to perform the planned spectral analysis, an
accurate determination of the pulse repetition period was
required. To obtain this, the average time increment, At,
between pulses (Table I) was computed. This increment was
the measure of the frequency with which an acoustic pulse
"sampled" the medium.
The time increment, At, was found by computing the total
time elapsed between the first sample on the first pulse in
a run, and the first sample on the last pulse in the same
run. This elapsed time, referred to as file length in
Table I, was then divided by one less than the total number
of pulse heights.
The analog-to-digital sampling period of 0.0*1 seconds
was the maximum error possible between the actual analog
file length and the digital file length as computed above.
Thus, for a file having a nominal 601 pulses (see Table I),
the maximum possible error in the computed average pulse time
increment, At, was on the order of:
(At) = - •°!ln^ec = ± 6.7 xlO"5 seconds
max 600
error
An error of this magnitude was considered acceptable for the
precision required to perform a valid spectral analysis of
the pu]se heights.
39
Computation of certain statistical parameters was the
next step in the analysis scheme. A pre-compiled subroutine
available in the Naval Postgraduate School IBM 360 library
was used to compute the coefficient of variation, the
minimum and maximum values and the range of the distribution.
The pulse height histogram was also computed and plotted.
It was immediately apparent from the histograms that a
small percentage of pulses had either extremely high values
or low values. However, this result was expected from the
Brush recordings which showed a random occurrence of
"drop-outs" and "spikes". The origin of these anomalies was
not exactly known, but it was apparent that they did not
represent real acoustic fluctuations. To eliminate these
a program was written to filter the pulse heights through a
gate whose width could be varied. The acceptable gate width
was determined by a trial and error process. The method
employed in this smoothing process was to compare a pulse
height with a running mean computed from the previous ten
pulses. The first ten pulses In a file were tested against
the mean of the first twenty heights to initiate the process,
The gate was centered around the mean. A bad value was
eliminated by substituting an artificial pulse whose height
was determined from a linear interpolation between the
preceding good pulse and the next succeeding good pulse.
Pulse heights filtered in each computer run were compared
with the Brush chart to determine if any good data had been
filtered or if any anomalies had been missed. This method
HO
did not eliminate all bad pulses and on some runs one or
two erroneous pulses escaped correction. Since these few
represented a very small percentage of the total sample —
less than 0.5$ — no further attempts were made to smooth
them. Thus the temporal fluctuations in the smoothed
distribution were, with rare exception, the result of actual
acoustic variability.
The smoothed data were then re-analyzed as discussed
above. Of all the basic statistics computed, the coefficient
of variation was the most pertinent. Skewness and kurtosis,
which were computed along with other statistics, would have
been valuable for comparison of histograms but were
invalidated for this purpose because the class interval
varied among distributions. In computing the histograms,
the constant factor was the number of class intervals rather
than the width of the class interval; this feature was a set
characteristic of the subroutine.
To facilitate comparison of results among runs, the
pulse heights of each run were normalized to the mean for
that run. Each pulse height sequence, then, had a mean of
1.0, and each pulse value was expressed as a percentage of
the mean. This process had no mathematical effect on the
coefficient of variation. Conversion of pulse heights to
standard variables was also done so that distribution in
terms of standard deviations was available.
Hi
G. SPECTRAL ANALYSIS
Following procedures detailed in Bendat and Piersol
(1966), preparation of the pulse height time series for
spectral analysis started with removal of the mean and of
any linear trend present in the series.
An estimate of the autocorrelation function, R(t), for
each time series was computed using the formula:
, N-m
R(t) = ± E (X)(X x )
N , n n+m
n=l
where N = the number of time samples, i.e., the
number of pulses in each run
m = the lag number
t = the lag time = m(At)
X = the individual pulse height value
The values for t were incremented from t=0 to a maximum lag
value of approximately one-tenth of the file length. The
increment used was the At determined by the method in
Section II. F.
A Parzen lag weighting function was then applied to the
autocorrelation function in order to insure a better estimate
of the spectral energy density values.
The autocorrelation function was Fourier transformed to
arrive at an estimate for the spectral energy density
function, G(f). The digital programming was derived from
the integral expression:
H2
G(f) = K I R(x) cos 2TTfT dx
0
where R(t) = the autocorrelation function
f = frequency
x = lag time
Estimates were computed for narrow frequency bands over the
range from D.C. to f where f is the Nyquist frequency
applicable to each run.
The cross correlation function R (x) between Hydrophone
xy
1 and Hydrophone 2 was computed next, using the equation:
n N-m
Rxy(T) = N=m" \ XnYn+m
n=l
where X = pulse height value for Hydrophone 1
Y = pulse height value for Hydrophone 2
and N, m, and x were as previously defined.
As with the autocorrelation functions, a Parzen window
was used to smooth the cross correlation function prior to
computation of the cross-spectral energy density function.
The co-spectral density and quadrature spectral density
2
functions were used to obtain the coherence, y , and
phase, 0 , functions for the pulse height time series:
2/ % c 2(f) + Qxy2(f)
xy G (f) G (f)
xv y
'13
-1 Qxv(f)
6 (f) = tan 1 SL
xy c (f)
xyv
2
where y (f) = coherence between Hydrophone 1 and
xy
Hydrophone 2 as a function of frequency
6YV(f) = phase between Hydrophone 1 and
xy
Hydrophone 2
C (f) = co-spectral energy density function
Q (f) = quadrature spectral energy density
function
G (f) = spectral energy density function for
Hydrophone 1
G (f) = spectral energy density function for
Hydrophone 2 .
H. DISPLAY OF RESULTS
The following computer generated plots were made on the
CALCOMP plotter:
1. relative pulse height time series for each hydrophone
2. autocorrelation functions for each hydrophone
3. auto-spectral energy density of the acoustic
fluctuations for each hydrophone
!\ . phase and coherence between Hydrophone 1 and 2 .
Except for one instance (Run 73) where the file length
was longer than usual, all runs were plotted using identical
scaling factors.
i\H
III. RESULTS OF ANALYSIS
A. TEMPERATURE FLUCTUATIONS
Detailed analysis of the temperature micro-structure data
was not performed in the course of this thesis. However,
the Brush chart recordings previously described afforded an
opportunity for a* qualitative evaluation of the micro-
structure activity present during each run. Of the twelve
runs under discussion, four showed temperature fluctuations
that were considerably more active than the others. The
remainder of the runs occurred under temperature conditions
varying from negligible to moderate activity. The most
active runs were numbers 72, 73, 116, and 117. Representative
samples of the twelve temperature records are shown in
Appendix A.
The temperature microstructure fluctuations were
evaluated using the thermistor calibration factors, tape
recorder gains, and the gain and chart speed settings for
the Brush recorder. An arbitrary classification scheme was
employed to assign qualitative descriptors:
Temperature Fluctuation Classification
< 0.05°C Negligible
.05°C to 0.2°C Moderate
> 0.2°C High
'15
B. STATISTICAL RESULTS OF PULSE HEIGHT ANALYSIS
Table III summarizes the statistical results obtained
from the analysis. Reference to Tables I and II supplies
information relating the acoustic frequency, time of day,
and environmental conditions to these results.
The coefficients of variation ranged from a low of 2.03$
(Run 56) to a high of 15.53% (Run 116) . The five highest
coefficients occurred in runs of relatively high thermal
activity. In general, the variation observed on Hydrophone 2
was slightly higher than that on Hydrophone 1, the exceptions
being Runs 45, 46, and 72. However, in only four runs (60,
73 » 116, 117) was the difference between the two hydrophones
greater than 2%; three of these four runs were also classified
as having higher microstructure content than the others.
Significant differences between the high frequency runs
and low frequency runs were more difficult to determine. For
runs that were conducted in high and low frequency pairs (see
Table I), the coefficients of variation for both Hydrophone
1 and 2 during the high frequency run were generally slightly
higher than during the low frequency run. However, since
the two runs in a pair were conducted 8 to 2*1 minutes apart
(Table I), the difference may not have been due to frequency
change alone.
Table III also has two sections pertaining to the range
of the distributions of pulse heights. The first of these
sections lists the extremes of the distribution in terms
of standard deviation. The second shows the range in terms
16
in % Relative to Mean
Maximum Range
c— C—
T~i co
co ^j-
o o
r-i CO
men
*
VO LTV
OJ VO
rH Ln
on t—
CO rH
r-i C—
CO O
inoj
o oo
on-=r
D— vo
on co
rH^T
OJ D—
VO CVJ
i>-cm
r-i tr~
rH rH
rH^r
co on
ooco
OJ OJ
OJ LO
O CT\
■=r o
OJ OJ
t— on
r-i C—
-=T CO
OJ OO
OO oo
oj-=r
LO OO
rH OJ
LO rH
in co
OOOJ
OJ -=r
OOVO
r-i O
oj^r
ooo
co co
OOOJ
OJ oo
inoj
^r in
onco
onco
r-i o
o vo
invo
co o
CO ON
moo
OCO
^r o
ON oo
-=r co
o t—
ONI>-
on r—
o o
rH rH
moo
r-i r-\
r-i r-i
rH O
rlrH
LO t»-
rH r-i
rH rH
co o
O r-i
r~i r-i
OOrH
r-i OJ
r-i r-i
r-i ON
<-i r-i
H rH
r-i l>-
r-i r-i
r-i rH
CO ON
OJ r-i
r-i r-i
r-iKD
OOOO
H H
VO o
r-i in
rH H
r— on
rH-=T
rH r-i
Distribution
Minimum
VO O
^T r-i
ON on
vo in
co i>-
t— co
OM—
VO O
ONOJ
r-i VO
CM OO
O l>-
O O
■=r oj
o ^r
-=r .=3-
r-i ON
f— in
CO r-i
t— co
ooin
t— o
o oo
ON ON
ON UN
CO CO
VO ON
co co
oco
ON [—
CM t--
ON CO
ON ON
CO l>-
O ON
ON t—
co in
co co
co o
l>-C—
rHVO
CO VO
l—O
C— VO
co vo
OJ ON
ON CO
ON r-i
on cr\
t>-vo
OJ r-i
VO oo
ONOJ
■=r oj
co vo
VO r-i
HVD
co oo
CO !>-
CM LO
moo
co en
rH-=T
CO CM
vo oo
on in
in t—
Cm a
O K
VO VO
m ON
t~- LO
LTM>-
[-- in
vo c—
vo in
VO LO
inin
oo i>-
vo in
vo in
n in Tern
aximum
-=3" CO
mvo
-=T OO
o on
OOVO
■^r vo
C— OJ
r-i^O
OJ LO
t— oo
OOON
oooo
i>-co
r-i r-i
on en
ON rH
r-i O
inoo
co oo
co oj
in Ln
OOOJ
en en
00^3-
OOOJ
oo oo
■=T OJ
oooo
OOOJ
oo oo
OJ CM
moo
oj on
oj en
Distributlo
Minimum M
CO r-i
m cm
LOCO
OO ON
OO OJ
ON LO
OJ c—
OJ ON
OOVO
VO O
OS CO
t—vo
ino
r-i ON
co oo
r-HVO
in o
CM OO
C--OJ
r-i in
r-i CM
on in
O OJ
r-i OJ
on on
1 1
OJ ^J"
1 1
OOOJ
1 1
OJ oo
1 1
OO OO
1 1
OJ oo
1 1
OOOJ
1 1
oocm
1 1
OJ oo
1 i
oooo
l l
OOOJ
1 1
-=r oj
l l
* — -*
Coeff of
Variation 1
l>-VO
t— ,-i
co oo
C\J CO
•=r t—
oo o
co .=3-
o oo
oooo
O OJ
co in
c— o
r-i O
inoj
vo in
r-HVO
VO ON
co o
t—LO
ON OO
vo in
OOOJ
vo on
OJ OJ
^r oj
oo-=r
■=r in
oj ^r
ooin
one-
on in
ON CO
in on
in in
r-i
t— in
r-i
# of Bad
Pulses
OJ OJ
r-i on
rH rH
r-i r-i
vo in
rH
O rH
l>-0O
r-i in
CD o
O r-i
rH on
VO oo
r-i
LOCO
OOON
in oo
OJ
CO ON
OJ
Hydro-
phone
r-i OJ
r-i OJ
t~\ OJ
r-i OJ
r-i OJ
r-i OJ
H OJ
r-i CM
r-i OJ
r-i OJ
rH OJ
r-i CM
It
s>a
SH
oo o
CT\H
-=r m
'in 1
co m
n . ;>:
O Q
v i J , !
^s
CM f-H
ooh
VO Q
,1,1
r 1
rH U
r-i
-p
s
■p
A
w
-p
r-i
W
0)
co
•H
m
>,
r-i
a
<
rH
a
o
•H
±>
W
•H
-P
a
-p
CO
Cm
o
>>
M
I
CO
H
H
« ii
■m 5
p- - >-
X X
O 0
□
1
-B-
O D
Q3 O
Q O-
n
in
o
0)
co
rH
-a
cti
o
0)
e
>
*^CN
00 G
IJJ o
_i '"
a. cti
•H
a fc
< cti
a >
IS. ^
O o
«
z
.o
-co
-P
c
o
•H
m
N CD
>j a,
rH 6
cti cd
d -p
<
c
•
-p o
>5
.C
P
bO C
•H
■H £
>
CU O
■H
W £
P
to
o
CD
cti
to T3
r-\ O
iH
3 -H
cd
CM U
E
CD
M
6 P.
CD
O
.c
M CD
p
Cm -P
3
0
e c
H
cd -H
,0
fc 6
•H
bO
bO
o m
•H
+s
rH
to a;
bO
•H £
:
O
l/l
CO
t—
-J
1/1
3
vy
b.
lil
£
CD
>>
.c -p
^> -H
>
bO-H
£ -P
•H
o
^
cd
3
T3 i-l
cd
•a
6
cd
f-i
-a
CD
^ &
o
p
o
cd ^
u
hO
•H
w £
>p
rH
cd
cd
U
c
CD
<
ft
E
p
CD
.c -p
M
•H
C
CD
o
m
a
0
£
w
O
rH .£
3
w
PM
■d
S
o
o
•H
u
?H
-.__ ,A
1 :.!:.":~
"..:;!(,
I".*! "t ■
(
|. ■:. I::-;::
■;">
I ■ j • | N
L- .::;•• ; • i
: ::;:::^ ; :
!-:
' ( . . ' '. J '..'.'
jr..-: i ■
E :f ;;. •■..
1
i.- f ..i ": ■:.
* ;-- • ■ ;
:':!.i,,i r
! : '■ " ;•
:\
j ! ;
}■■■ \
I : .
I >
Ni-
>3
-P
•H
>
•H
P
O
ctf
£>
•H
b0
•H
rH
(D
o
•H
■P
OJ
P
o
0)
u
P
e
Eh
oo
c
K
73
-- U-
■f>:|:
M~
■J:.-: .•■":-; [
l:.::
".'..,:'..'". (.■. i. 1
!
t —
~i "li": ;'"■;■•■•?
.■i""ft:--r-
lEii
■~i:vj:.::-; -:-v:- $"
.:; : ;;:-[ :
t ---
^fiia
••-;: :.| ..[.:• 'J \ ■-:;.-..;:_.;.; - :: ~i
r_ J"- "■■;£" "":: "
I] IS^L.
; :y
f "■!-. .::;:;
!.:;|
[£•:• :::: : ':: '
tJML
:::--]
F" ' -'--;- -
- j :1 L : : •":
■ - -I
:l " ■!
7*J
:iia
FFWB
i.:::l'
|. -:-r--
t- y
'm
r-l-
I'LL:!
I ■ !■ ! .
-P
■H
>
•H
-P
O
CtS
CD
P
'cd
M
CD
T)
O
w
o
•H
P
cd
-P
o
iH
Cm
CD
U
P
ctf
M
CD
s
CD
Eh
o
vo
C
>>
P
•H
>
•H
P
O
Cti
CD
P
nJ
M
CD
o
ra
C
o
•H
P
Cd
P
O
3
CD
M
P
cd
(II
a
6
CD
Eh
C\J
K
75
r ...,....;
:i TFT^]
1
P>
•H
i
>
•H
-P
O
ctf
t r — r -\
£1
/ KH
i
/
.^
>
u
•p
ctf
a
ro
i !
T
i:! ^4
Btl
:H
.^_T+
■-£-----
li^l
I - !
... ;. ..:
£.
^H
VD
K
f
i i ! I
: . . i
■'. ..; i
1_ -— ~
-(N-71
9 I
\
>>
-p
•rH
>
•H
P
O
Ctf
(D
P
Ctf
Jh
0
>cs
o
6
10
C
o
■H
P
ctf
P
o
H
Cm
Q)
p
ctf
U
0>
(X
E
c
76
-co
ra
-p
M
•H
CD
0)
to
-m
3
"5
-n
-CN
in
O
d
O
O
O
NV3W Ol 3AI1V1?1U S1H9I3H 3Sind
faO
•H
(D
0
w
cH
PL,
0
o
(X
o
-Ci
n i i •
fM
-c>*
o
d
-P
Si
bO
•H
(U
Sd
O)
CO
.H
a+
C\J
>
K
C
KV3W 01 3AI1V13U S1H9I3H 3Sind
8?
a
o
■H
-P
O
C
c
o
■H
-P
cd
>
-ro
oo
-a
l-
in
o
1^
CM
o
cf
o
6
NV3W 01 3AllV13cJ S1M9I3H 3S"ind
86
M
c
o
•H
■P
O
c
o
•H
-P
:
3
Ph
m
m
iH
2
(I)
F-
£
O
Xi
■<*
o.
u
^
•c
>>
w
n
OA
-CN
O
O
NV3W 01 3 AI1V13JJ S1HDI3H 3Sind
89
I
o
o
1^
•o
oo
-t^
M
e
•*•
3
C
2
-*■
-n
o
6
NV3W oi 3Aiivi:y snioinn 3sino^
to
-p
bQ
•H
*
a;
w
0
O
O
U
-CI
^g
€
-CN
r-
in
5
o
1^
d
o
d
o
d
NV3W OJ. HAI1V13H S1H9I3H 3Sind
93
-o
-CO
-t»
-° ^.
-p
5
96
-o
-co
w
-p
£1
60
•H
* Q)
3 CO
CrH
^3
Ul
± C
»- o
o
-ft
CO
in
-CN
r-
CM
o
6
ri
O
O
6
NV3W Ol 3AI1V13H S1H9I3H 3Sind
97
-o-
-00
o
d
d
O
d
NV3W 01 3AIIV13U S1H9I3H 3SlHd
98
Kl
■P
2
-T
CO
2 3
0
c
o
.a
o
-a
-ro
CO
in
-•
Z
UI
Q
-4.50
\
<
1—
o
UI
a.
CO
O
O
-5.25
-6.00
—
' *HP 2
i -
I I J.
0.00
0.15
0.30
0.45
0.60
0.75
1.0,-
-100
0.75
0.7&
FREQUENCY (Hz)
]00
-o>
-co
-t»-
w
■p
.G
bO
•H
(i
«- QJ
3 CO
C H
5 5
2
-*t
CD
c
O
.G
ft
O
Ph
>3
-co
o
-CN
j?
o
•o
o
o
6
NV3W 0.1 3AI1V13H S1H9IHH 3S10d
101
-co
-t*
-O ^.
CD
-in
ui
2
-t
0
ra
2 3
CM
d>
o
o
u
•m
o
U3
c
3
K
-CN
in
1^
o
6
O
O
O
NV3W Oi 3AI1VT3H S1MDI3H 3SlfUl
102
K>
C
o
•H
-P
O
c
O
•H
-P
cd
rH
•
z
Ul
o
-4.50
-J
<
u
UJ
a.
«/>
O
-5.25
\l/^ \ /\HP1
\ /^~\ ^■v^^T'Y^^ /HP 2
-1
-6.00
(
).00
1
0.15
0.30 0.45 0.60 0.'7£
10r-
UJ
U
z
iu
tc
ut
X
o
u
0.0
0.00
0.15
-^
0.30
0.45
0.60
0.75
+180r-
Ui
<
a:
CL
-100I-
FRCQULNCY (Hz)
0.75
108
-o.
-co
"° *-»
w
■P
w
•H
0
K
-m
ui
2
-t
-m
-CM
CO
2 3
^ Pm
c
o
a
o
-a
CM
£
3
K
"nr-
o
o
in
n
d
o
6
NV3W Ol 3AI1V13H S1H9I3H 3S1Dd
Hi'j
r~
o-
o
o
d
m
d
-00
CO
•p
.C
h0
■H
n
CD
O
CD
3
w
C
fH
2
3
dn
in
CM
UJ
2
CD
-t
-ro
C
o
Xi
o.
o
X)
CM
CM
O
d
NV3W Oi 3AI1V13U S1H9I3H dS'lDd
110
M
c
o
•H
-P
O
c
c
o
•H
•P
cti
H
0)
O
o
o
-p
c
NOIiDNOd N0liV13a«0D0inV
111
1.0
-1.0
Run 72. Spectral Density,
Coherence, and Phase
z
iu
Q
<
V-
O
Ul
a.
©
O
-3.0
-5.0
-7.0
0.00
0.15
0.30
0.45
0.60
0.75
iu
U
z
ui
Ui
3C
O
u
10r-
+180r-
Ul
1/1
<
X
0.
-1801-
FREQULNCY (Hi)
112
1-°
"Sr
CO
-
Run 73- Spectral Density,
Coherence, and Phase
0.00
0.15
0.30
0.45
0.60
0.75
U
z
ui
de:
UI
X
O
o
1.0r-
0.0
yVW A ^/^^/w. AA^An.^/
0.00
0.15
0.30
0.45
0.60
0.75
+180r-
<
X
-1001-
FREQUmCY {\\z)
0.75
116
-©>
-eo
-P
£1
bO
•H
0
Kl
-O^
-m
2
-•<*•
c
o
.£
o
u
>>
-co
r-l
-CN
m
I
o
o
1^
o
o
o
6
NV3W 01 3AI1V13H SIII0I3H 3Sind
117
-c
-oo
-t*
w
-P
-O ^
0)
n
c
-in
2
-**
to
■H
P-,
OJ
C
o
o
-a
-TO
-CM
O
b
0
~7_
O
O
6
NV3V< Ol 3AI1V13U SIH9I3H 3S"Ifld
118
CO
C
O
■H
■P
O
C
C
o
■H
•P
cd
iH
0)
U
O
o
o
•p
VD
r-\
iH
NOiiDNnd NOiivnayyoDoinv
119
-2.0
-3.0
Z
<
DC
I-
o
flu
t/J
O
O
-4.0
-5.0
Run 116. Spectral Density,
Coherence, and Phase
-6.0
0.00
0.15
0.30
0.45
0.60
0.75
10r-
ui
O
Z
u
tu
m
X
O
u
0.0
0.00
0.15
0.30
0.45
0.60
0.75
+180r-
IU
<
X
a.
-100
FREOUCMCY (Hz)
0.75
120
-t>
-00
-t«.
CO
-p
-o*-.
— CO
i 3
-in
2
-*r
c
o
o
u
-M
iH
iH
C
d
-CN
i
o
o
1^
•o
"T-
o
o
O
NV3W 01 3AliV13cJ S1H0I3H HSlHd
121
— r
o
1^
o
o
CO
-m
CO
-p
.c
M
■H
0
ffi
M
o
0)
••
m
3
rH
c
3
2
Ph
CM
Ul
CI)
i
£
-co
-CN
- -o
o
6
NV3W Oi 3AI1V13
Michael Thomas Korbet
t. CONTRACT OR GRANT NUMBERS
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Naval Postgraduate School
Monterey, California 93940
10. PROGRAM ELEMENT. PROJECT, TASK
AREA 4 WORK UNIT NUMBERS
I. CONTROLLING OFFICE NAME AND ADDRESS
Naval Postgraduate School
Monterey, California 93940
12. REPORT DATE
March 1974
13. NUMBER OF PAGES
131
14. MONITORING AGENCY NAME & ADDRESS!-// dlllerent from Controlling Office)
Naval Postgraduate School
Monterey, California 93940
IS. SECURITY CLASS, (of thlt report)
Unclassified
I5«. DECLASSIFI CATION/ DOWN GRADING
SCHEDULE
16. DISTRIBUTION STATEMENT (of this Report)
Approved for public release; distribution unlimited.
17. DISTRIBUTION STATEMENT (of the abttrect entered In Block 20, If different from Report)
IB. SUPPLEMENTARY NOTES
Research supported through contracts from
Naval Ordnance Systems Command (Code 03C)
19. KEY WORDS (Continue on reverie tide If necetttry and Identity by block number;
Underwater acoustics
Shallow water acoustics
Acoustic amplitude fluctuations
Temperature microstructure
20. ABSTRACT (Continue on reverie tide II necetttry and Idtnttty by block number)
An underwater acoustics experiment conducted in shallow water
(70 feet) off the New Zealand east coast in 1972-1973 is
described. Short acoustic pulses of 35 and 65 kHz sound were
projected along near-orthogonal paths of approximately 300 yards,
Environmental parameters were simultaneously observed.
Statistical and spectral analyses of pulse heights were
performed on 12 selected runs using digital techniques.
DD ,^73 1473
(Page l)
EOITION OF I NOV 65 IS OBSOLETE
S/N 0 102-014- 660 1 I
130
UNCLASSIFIED
SECURITY CLASSIFICATION OF THIS PAGE (When Dele Bnltitri)
UNCLASSIFIED
4'bt.URlTY CLASSIFICATION OF THIS PAGEfWim Dmlm Enttrtd)
(20. ABSTRACT continued)
Coefficients of variation ranged from 2.0$ to 15 - 5% - In almost
all cases, higher variability was observed along the acoustic
path oriented perpendicular to the predominant swell direction.
Along this same path, periods corresponding to common surface
swell periods were frequently evident in the autocorrelation
functions of the fluctuations. Coherence between the fluctuations
along each path was low, averaging about 0.1. Long period oscil-"'
lations suggestive of modulation by internal waves were apparent
in several runs. Mo significant dependence of variability on
acoustic frequency was detected.
Microscale temperature fluctuations measured simultaneously
are discussed.
DD Form 1473 (BACK)
1 Jan 73 . UIICI ASSTF1FP
S/N 01 02-014-()()(> 1 SECURITY CLASSIFICATION OF THIS PAGF.fH7..n O.r. F.nlmfd)
■i J 1 U b I
Thesis
K786 Korbet
am^f!l°Wwater acoustic
amplitude fluctuations
at J5 and 65 kHz.
151061
I Thesis
i;786
fc.l
Korbet .
Shallow water acoust-
amplitude fluctuations
at 35 and 65 kHz.
thesK786
Shallow water acoustic amplitude fluctua
3 2768 002 10491 1
DUDLEY KNOX LIBRARY